pulse note desk

ServiceNow down 30%+ YTD in its worst quarter: AI is eating SaaS, except when it isn't

The stock joined the $1T software rout. Customers did the opposite.

You watched ServiceNow drop over 30% year-to-date in 2026—its worst quarter on record—while the company guided 20% subscription revenue growth for the full year and authorized another $5 billion in share repurchases. Consensus believes AI agents from OpenAI, Anthropic, and hyperscalers will commoditize enterprise workflows, turning ServiceNow's high-margin subscription seats into low-value API calls. That's why software valuations cratered nearly $1 trillion and NOW trades near 52-week lows despite beating estimates. The numbers tell a different story: customers are reloading credits on Now Assist, renewing at 98%, expanding workflows, and signing bigger deals. The selloff prices in disruption that the backlog and consumption data have not shown up yet.

Q4 2025 subscription revenue hit $3.466 billion, up 21% year-over-year and 19.5% in constant currency, beating guidance by 1.5 points per the earnings release. That came on top of full-year 2025 subscription revenue of $12.883 billion, also up 21%. You're looking at durable double-digit growth off a much larger base while the broader SaaS basket got repriced on AI fear. The market sold the fear. Customers bought the platform.

Current remaining performance obligations reached $12.85 billion, up 25% year-over-year (21% constant currency). That backlog gives you clear line of sight into the next twelve months of revenue and shows customers committing more upfront, not less. When cRPO grows faster than current revenue, it typically signals accelerating demand. Here it did exactly that—25% versus 21%—while the stock traded as if the business faced evaporation risk.

Renewal rates held steady at 98%. Nearly 3,000 customers adopted Now Assist, with Assist pack consumption up 55x since last May. Workflows and transactions each expanded more than 33%, and the company logged 244 deals over $1 million in net new annual contract value—nearly 40% growth year-over-year. Customers aren't pausing to test whether agents will replace them. They're layering AI automation on top of the existing control tower for IT, HR, and customer service workflows. Now Assist isn't a replacement threat—it's an expansion engine inside the platform they already trust.

The board authorized an additional $5 billion share repurchase program in January 2026, including a planned $2 billion accelerated buyback, on top of prior capacity. Management is putting real capital to work to offset dilution and buy shares at these depressed levels. You don't authorize that scale of buyback if you expect the core subscription model to commoditize into thin-margin API traffic. The move signals confidence in cash generation and long-term durability.

Valuation compressed sharply. The forward P/E sits in the mid-20s after trading north of 50x not long ago. You're paying less for each dollar of guided growth precisely because sentiment detached from the metrics. Early AI adopters are stocking up on credits today. Pandemic-era contracts come up for renewal later this year, providing another visible tailwind. The setup points to steady, high-visibility expansion at improving margins—non-GAAP operating margin expanded to 31% in Q4 and guidance calls for 32% in 2026.

ServiceNow isn't fighting AI disruption. It is the AI control tower that makes agents useful inside large enterprises. Now Assist sits on top of the deterministic workflows customers already run on the platform. Switching costs are massive. Consumption is accelerating. The deadpan fact bomb: ServiceNow guided roughly 20% subscription revenue growth for 2026—to $15.53–15.57 billion—and authorized another $5 billion buyback. The stock still dropped 30%+ YTD because someday, somehow, autonomous agents might replace seats. Customers are currently buying more seats, more credits, and more workflows.

This is not blind optimism. The thesis has clear, measurable tripwires. If 2026 subscription revenue growth falls below 18% in constant currency at the next guidance update, the growth narrative weakens. If cRPO growth decelerates below 20% year-over-year in the Q1 or Q2 2026 print, visibility into demand is cracking. A renewal rate drop below 95% or material deceleration in net new ACV in the first half would signal customers pulling back. Gross margin breaking below 80% for two consecutive quarters on AI-related costs would question profitability durability. And if hyperscaler or custom AI agents capture more than 15% share of enterprise workflow spend by end-2026 per customer surveys or public filings, the control-tower advantage narrows.

None of those signals have appeared. Instead you see 21% subscription growth, 25% cRPO expansion, 98% renewals, exploding Now Assist consumption, and aggressive capital return. The market priced the fear. The data keeps delivering the reload. ServiceNow checks every box for a durable enterprise platform growing 20% with sticky renewals and accelerating AI usage layered inside it—while the multiple has reset to something far more reasonable.

You don't need to chase the next standalone AI hype cycle to make money here. You need a platform that large enterprises already depend on, with visible backlog, high renewal rates, and AI driving incremental spend rather than substitution. ServiceNow delivers that today. If execution holds and the kill criteria stay quiet, patience here gets rewarded. The crowd that sold first may end up watching the recovery from the sidelines.

The punchline is straightforward. AI isn't eating ServiceNow. It's feeding it. The stock simply forgot to read the invoice.